🚀 go-pugleaf

Muke Schuster and others have written about setting drive speeds, et
c., and the fact that strobe wheels are a a good deal more accurate
than DSPEED.TOS or whatever.  Fine.  But there has been an unspoken
assumption all along, and that is:  You want your drive to be going
_exactly_ 300 RPM.  I question this.  Yes, that's the specification
alright, but there is a good reason to run your drives a little bit
SLOWER (say, 297 RPM or so).

The reason is thusly:  if you are using one of those clever
formatters, such as DCFORMAT or any of its ilk, that can put 10
sectors on a track, the last gap (Gap #4, made out of 0x4e's) gets a
little bit shorter than normal, which is 1401 bytes; it gets trimmed
down to 780-odd bytes.  It's not a bad idea, however, to run slightly
larger than normal Gap 3's (normally 40 bytes, I like to double
this), and then Gap 4 really gets crowded.  But if you run the drive
a little slow, the effect is that to the 1770, all the tracks look a
little longer than normal, and you get your Gap 4 size back to
something reasonable.  You can check this by looking at the data
coming off the read line with an oscilliscope (take the scope sync
input from the index pulse line) and noting how close the end of the
last sector comes to the end of the track.

I have done this routinely for years and find it doesn't generate
any trouble if you are only 1 or 2% slow; the PLL in the data
separator can handle this kind of an error with it's eyes closed,
and the added security of the longer gaps helps to overcome all
kinds of write-skew and drive wow-and-flutter problems.

Ross Alexander @ Athabasca University,
alberta!auvax!rwa